Applied AI, described through the decisions it improves.

An AI-assisted root-cause and workflow system that helps planners understand supplier performance and move from investigation to action.

Decision problem

Supplier exceptions required planners to assemble evidence across signals, identify chronic patterns, quantify impact, and prepare vendor communications manually.

Approach

Combined machine-learning signals, structured root-cause analysis, and agentic workflows with human review at the decision boundary.

Outcome

Reduced root-cause analysis effort by 5–10× and planner communication effort by approximately 80%.

Risk models that focus inspection capacity on the distribution-center days and purchase-order lines most likely to need attention.

Decision problem

Fresh-quality teams need to direct limited inspection time toward the highest-risk inventory without losing coverage of rejected items.

Approach

Used supplier, purchase-order, claims, throws, returns, product-age, recency, and inspection signals in gradient-boosted risk models.

Outcome

Backtests captured approximately 80% of rejected items within a smaller, higher-risk inspection set.

Mixed-integer optimization systems that translate operating constraints into practical allocation, logistics, and workforce decisions.

Decision problem

Allocation and labor decisions involve intertwined capacity, service, transportation, and workforce constraints that heuristics handle poorly.

Approach

Formulated MILP models for meat allocation and less-than-truckload logistics, as well as part-time and full-time labor planning.

Outcome

The meat allocation solution produced approximately $8M in annual savings; labor optimization identified approximately $5M in potential savings.

A discrete-event simulation that makes complex container-flow and capacity trade-offs visible before operational decisions are made.

Decision problem

Renovations, disruptions, dray capacity, storage limits, and container backlogs interact in ways that static planning models cannot represent well.

Approach

Built a Python and SimPy model of container flow, dray capacity, backlog, storage utilization, and demurrage and detention exposure.

Outcome

Enabled planning teams to evaluate capacity, renovation, and disruption scenarios in a controlled decision environment.